Color code: (critical, news during the meeting: green, news from this week: blue, news from last week: purple, no news: black)
High priority Framework issues:
- Fix dropping lifetime::timeframe for good: Still pending: problem with CCDB objects getting lost by DPL leading to "Dropping lifetime::timeframe", saw at least one occation during SW validation.
- Newly spotted bogus message about decreased oldestPossible counter should be suppressed. Status?
- Start / Stop / Start: 2 problems on O2 side left:
-
- All processes are crashing randomly (usually ~2 out of >10k) when restarting. Stack trace hints to FMQ. https://its.cern.ch/jira/browse/O2-4639
- TPC ITS matching QC crashing accessing CCDB objects. Not clear if same problem as above, or a problem in the task itself:
- Stabilize calibration / fix EoS: New scheme: https://its.cern.ch/jira/browse/O2-4308: Status?.
- Fix problem with ccdb-populator: no idea yet - since Ole left, someone else will have to take care.
- Need to reduce metric intervals for Pb-Pb with all EPNs in. Made --resource-monitoring configurable in O2DPG. Need to do similar for DPL metrics.
- Would propose to add a command line option to raise the interval of all metrics to a configurable minimum interval.
Global calibration topics:
- TPC IDC and SAC workflow issues to be reevaluated with new O2 at restart of data taking. Cannot reproduce the problems any more.
Sync reconstruction
- Waiting for RC to test COSMIC replay data set.
- STOP timeout discussion: RC did some tests, which shows a significant dependence on the STOP timeout. However only tested in SYNTHETIC. Should repeat in PHYSICS.
- CTF count monitoring available now. Ernst had a look, and for most runs we lack 300-800 CTFs, i.e. 1-3 seconds (at 20s STOP timeout).
Async reconstruction
- Remaining oscilation problem: GPUs get sometimes stalled for a long time up to 2 minutes. Checking 2 things:
- does the situation get better without GPU monitoring? --> Inconclusive
- We can use increased GPU processes priority as a mitigation, but doesn't fully fix the issue.
- ḾI100 GPU stuck problem will only be addressed after AMD has fixed the operation with the latest official ROCm stack.
- Limiting factor for pp workflow is now the TPC time series, which is to slow and creates backpressure (costs ~20% performance on EPNs). Enabled multi-threading as recommended by Matthias - need to check if it works.
- Problem reported by Max/Ruben that analysis fails, since sometimes at EOR we receive TFs with no data and bogus orbit.
- Giulio will implement tf-status message (from https://github.com/AliceO2Group/AliceO2/pull/13495) sent by readout-proxy.
EPN major topics:
- Fast movement of nodes between async / online without EPN expert intervention.
- 2 goals I would like to set for the final solution:
- It should not be needed to stop the SLURM schedulers when moving nodes, there should be no limitation for ongoing runs at P2 and ongoing async jobs.
- We must not lose which nodes are marked as bad while moving.
- Interface to change SHM memory sizes when no run is ongoing. Otherwise we cannot tune the workflow for both Pb-Pb and pp: https://alice.its.cern.ch/jira/browse/EPN-250
- Lubos to provide interface to querry current EPN SHM settings - ETA July 2023, Status?
- Improve DataDistribution file replay performance, currently cannot do faster than 0.8 Hz, cannot test MI100 EPN in Pb-Pb at nominal rate, and cannot test pp workflow for 100 EPNs in FST since DD injects TFs too slowly. https://alice.its.cern.ch/jira/browse/EPN-244 NO ETA
- DataDistribution distributes data round-robin in absense of backpressure, but it would be better to do it based on buffer utilization, and give more data to MI100 nodes. Now, we are driving the MI50 nodes at 100% capacity with backpressure, and then only backpressured TFs go on MI100 nodes. This increases the memory pressure on the MI50 nodes, which is anyway a critical point. https://alice.its.cern.ch/jira/browse/EPN-397
- TfBuilders should stop in ERROR when they lose connection.
- Allow epn user and grid user to set nice level of processes: https://its.cern.ch/jira/browse/EPN-349
- Tentative time for ALMA9 deployment: december 2024.
Other EPN topics:
Full system test issues:
Topology generation:
- 2 Occucances where the git repository in the topology cache was corrupted. Not really clear how this can happen, also not reproducible. Was solved by wiping the cache. Will add a check to the topology scripts to check for a corrupt repository, and in that case delete it and check it out anew.
AliECS related topics:
- Extra env var field still not multi-line by default.
GPU ROCm / compiler topics:
- ROCm 6.2.2 available from AMD, with should contain fixes for all the bugs we have reported with 6.x.
- EPN provided several test nodes. Unfortunately, it is again crashing on MI100 EPNs (MI50 EPNs are stable) with 2 independent failures:
- FST and standalone test is crashing after ~10 minutes with GPU memory error.
- This error goes away when I serialize all kernels and DMA transfers. Verified however that it is a different serialization problem compared to what we had 2 years ago, the workaround from back then does not work.
- As a side effect, after this error has appeared once, the node seems to become unstable and must be rebooted, or it will show strange behavior.
- As a workaround, we can serialize all kernels / DMA transfers, which reduces the MI100s roughly to MI50 performance.
- Servers freeze and reboot after some time of running.
- So far seen only in FST, not standalone benchmark.
- From last message to console before the freeze, seems to be related to shared memory registration to GPU, which could explain the error in the FST, where we register much more memory than in standalone tests.
- No workaround possible so far.
- Tried with same memory settings as on MI50 (thinking it could be the amount of memory registered), but it still crashes.
- Plan now is to push AMD to fix the freezes. If they succeed, we’d propose to update anyway to new OS / driver / runtime to move ahead. We’d need to serialize all kernels on MI100 until the memory error is fixed as well.
- New miscompilation for >ROCm 6.0
- Waiting for AMD to fix the reproducer we provided (not yet fixed in 6.2.2, but we have a workaround).
- Try to find a better solution for the problem with __device__ inline functions leaking symbols in the host code.
- Created PR to bump GCC to 14.2 (just to test).
- Fails in ROOT compilation for AliRoot: AliRoot switched to default ROOT, but CI must rerun?
- Fails to compile clang on SLC9 for ARM: Status?
TPC GPU Processing
- WIP: Use alignas() or find a better solution to fix alignment of monte carlo labels: https://its.cern.ch/jira/browse/O2-5314
- Waiting for TPC to fix bogus TPC transformations for good, then we can revert the workaround.
- Waiting for TPC to check PR which uses full cluster including average charge and occupancy map errors during seeding.
- TPC asked to be able to apply cluster cuts also at CTF decoding, not only during clusterization. Will add this.
TPC processing performance regression:
- Final solution: merging transformation maps on the fly into a single flat object: Still WIP
General GPU Processing
- Switched entire GPU code in O2 to using <cstdint> int types, which should fix some more issues with signed vs unsigned char on ARM.
